Echo

Privacy-first iOS voice assistant. Self-hosted LLM and TTS.

Swift SwiftUI Speech Recognition GPT-4 OSS 20B Kokoro-TTS AVFoundation

Overview

Echo is an iOS voice assistant that integrates modern AI for natural, real-time conversations. Built with Swift and powered by self-hosted models, it provides an intuitive voice experience that feels responsive and intelligent.

Key Features

Real-time Voice Recognition

Apple's native Speech Recognition framework for accurate, on-device speech-to-text with automatic silence detection.

Intelligent Conversations

Powered by GPT-4 OSS 20B for contextual, natural language responses that understand context and provide meaningful interactions.

High-Quality Voice Synthesis

Kokoro-TTS for expressive, human-like speech generation with customizable voices and speed.

Privacy-First Architecture

Both the LLM and TTS are self-hosted. Echo works entirely offline without sending any data to the internet.

Streaming Audio

Real-time audio streaming with immediate response playback as content is generated for responsive interaction.

Real time voice interaction demo

How It Works

1 Listen — Tap to activate voice recognition with visual feedback

2 Process — Speech is transcribed using Apple's Speech Recognition

3 Generate — GPT-4 creates contextual responses in real-time

4 Synthesize — Kokoro-TTS converts text to natural-sounding speech

5 Stream — Audio plays immediately as it's generated